| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
feat(mcore): add GLM-5.2 A5 POC scripts Co-authored-by: LinShua<707894133@qq.com> # message auto-generated for no-merge-commit merge: !5013 merge master_glm52_A5_sh into master feat(mcore): add GLM-5.2 A5 POC scripts Created-by: LinShua Commit-by: LinShua Merged-by: ascend-robot Description: Fixes #1825 ## What this PR does / why we need it? 1. Replaces the existing GLM-5.2 32B A5 POC entry with a 51B 4K FP8 configuration using TP1/PP2/EP4 parallelism, Transformer Engine MXFP8, index-topk=1024, fixed routing, the pipeline layout, and a matching log name. 2. Adds a GLM-5.2 210B 4K A5 FP8 POC entry with an eight-node TP1/PP2/EP32 configuration, Transformer Engine MXFP8 options, and portable data/tokenizer/checkpoint path placeholders. 3. Aligns --lr-warmup-iters to 500 in the GLM-5.2 744B A3 example, 744B A3 POC, and 210B A3 POC scripts. 4. This change is limited to five shell launch configurations and does not modify training runtime code or public APIs. ## Does this PR introduce any user-facing change? Yes. Users can run the new tests/poc/glm52/pretrain_glm52_210b_4k_A5_fp8_ptd.sh entry and the renamed tests/poc/glm52/pretrain_glm52_51b_4k_A5_ptd.sh entry. The 51B entry now uses the A5 MXFP8 and TP1/PP2/EP4 configuration, and the related A3 scripts now default to 500 learning-rate warmup iterations. ## How was this patch tested? 1. bash -n passed for all five changed shell scripts. 2. Changed-file pre-commit passed trailing-whitespace, end-of-file, added-large-file, merge-conflict, private-key, codespell, and typos checks. YAML, JSON, Python, and C/C++ hooks had no matching files. The repository-local gitleaks-offline-scan hook could not run because the required ./gitleaks executable is not present in the repository or local environment. 3. git diff --cached --check passed before amend. 4. The remote Git hook passed during git push --force-with-lease. 5. Functional, accuracy, performance, and end-to-end NPU training tests were not run. See merge request: Ascend/MindSpeed-LLM!5013 | 7 天前 | |
feat(mcore): add GLM-5.2 A5 POC scripts Co-authored-by: LinShua<707894133@qq.com> # message auto-generated for no-merge-commit merge: !5013 merge master_glm52_A5_sh into master feat(mcore): add GLM-5.2 A5 POC scripts Created-by: LinShua Commit-by: LinShua Merged-by: ascend-robot Description: Fixes #1825 ## What this PR does / why we need it? 1. Replaces the existing GLM-5.2 32B A5 POC entry with a 51B 4K FP8 configuration using TP1/PP2/EP4 parallelism, Transformer Engine MXFP8, index-topk=1024, fixed routing, the pipeline layout, and a matching log name. 2. Adds a GLM-5.2 210B 4K A5 FP8 POC entry with an eight-node TP1/PP2/EP32 configuration, Transformer Engine MXFP8 options, and portable data/tokenizer/checkpoint path placeholders. 3. Aligns --lr-warmup-iters to 500 in the GLM-5.2 744B A3 example, 744B A3 POC, and 210B A3 POC scripts. 4. This change is limited to five shell launch configurations and does not modify training runtime code or public APIs. ## Does this PR introduce any user-facing change? Yes. Users can run the new tests/poc/glm52/pretrain_glm52_210b_4k_A5_fp8_ptd.sh entry and the renamed tests/poc/glm52/pretrain_glm52_51b_4k_A5_ptd.sh entry. The 51B entry now uses the A5 MXFP8 and TP1/PP2/EP4 configuration, and the related A3 scripts now default to 500 learning-rate warmup iterations. ## How was this patch tested? 1. bash -n passed for all five changed shell scripts. 2. Changed-file pre-commit passed trailing-whitespace, end-of-file, added-large-file, merge-conflict, private-key, codespell, and typos checks. YAML, JSON, Python, and C/C++ hooks had no matching files. The repository-local gitleaks-offline-scan hook could not run because the required ./gitleaks executable is not present in the repository or local environment. 3. git diff --cached --check passed before amend. 4. The remote Git hook passed during git push --force-with-lease. 5. Functional, accuracy, performance, and end-to-end NPU training tests were not run. See merge request: Ascend/MindSpeed-LLM!5013 | 7 天前 | |
feat(mcore): add GLM-5.2 A5 POC scripts Co-authored-by: LinShua<707894133@qq.com> # message auto-generated for no-merge-commit merge: !5013 merge master_glm52_A5_sh into master feat(mcore): add GLM-5.2 A5 POC scripts Created-by: LinShua Commit-by: LinShua Merged-by: ascend-robot Description: Fixes #1825 ## What this PR does / why we need it? 1. Replaces the existing GLM-5.2 32B A5 POC entry with a 51B 4K FP8 configuration using TP1/PP2/EP4 parallelism, Transformer Engine MXFP8, index-topk=1024, fixed routing, the pipeline layout, and a matching log name. 2. Adds a GLM-5.2 210B 4K A5 FP8 POC entry with an eight-node TP1/PP2/EP32 configuration, Transformer Engine MXFP8 options, and portable data/tokenizer/checkpoint path placeholders. 3. Aligns --lr-warmup-iters to 500 in the GLM-5.2 744B A3 example, 744B A3 POC, and 210B A3 POC scripts. 4. This change is limited to five shell launch configurations and does not modify training runtime code or public APIs. ## Does this PR introduce any user-facing change? Yes. Users can run the new tests/poc/glm52/pretrain_glm52_210b_4k_A5_fp8_ptd.sh entry and the renamed tests/poc/glm52/pretrain_glm52_51b_4k_A5_ptd.sh entry. The 51B entry now uses the A5 MXFP8 and TP1/PP2/EP4 configuration, and the related A3 scripts now default to 500 learning-rate warmup iterations. ## How was this patch tested? 1. bash -n passed for all five changed shell scripts. 2. Changed-file pre-commit passed trailing-whitespace, end-of-file, added-large-file, merge-conflict, private-key, codespell, and typos checks. YAML, JSON, Python, and C/C++ hooks had no matching files. The repository-local gitleaks-offline-scan hook could not run because the required ./gitleaks executable is not present in the repository or local environment. 3. git diff --cached --check passed before amend. 4. The remote Git hook passed during git push --force-with-lease. 5. Functional, accuracy, performance, and end-to-end NPU training tests were not run. See merge request: Ascend/MindSpeed-LLM!5013 | 7 天前 | |
feat(pytorch): add GLM-5.2 51B A5 BF16 training example Co-authored-by: LinShua<707894133@qq.com> # message auto-generated for no-merge-commit merge: !5032 merge master_glm5.2_16die_A3_example_sh_0909 into master feat(pytorch): add GLM-5.2 51B A5 BF16 training example Created-by: LinShua Commit-by: LinShua Merged-by: ascend-robot Description: Fixes #1844 ## What this PR does / why we need it? 1. Add a GLM-5.2 51B 4K A5 BF16 pretraining example with configurable data, tokenizer, checkpoint, and distributed launch paths. The example uses the local Transformer implementation, TP2/PP4/EP2, DSA index-topk=2048, pipeline layout, and recomputation settings as a reusable POC entrypoint. 2. Add --disable-gloo-group to the DeepSeek3 60B 4K 128-card A3 pretraining script so that this NPU/NCCL launch path does not create a Gloo process group. 3. Keep the change limited to these two shell entrypoints. The broader script-wide Gloo configuration work remains tracked by #1828. ## Does this PR introduce any user-facing change? Yes. Users can launch the new BF16 example from tests/poc/glm52/pretrain_glm52_51b_4k_A5_ptd_example.sh after setting the data, tokenizer, and checkpoint paths. The existing DeepSeek3 60B A3 entrypoint now passes --disable-gloo-group by default. ## How was this patch tested? 1. bash -n passed for both changed shell scripts using Git for Windows Bash. 2. git diff --check origin/master...HEAD passed. 3. The repository changed-file pre-commit run passed trailing whitespace, end-of-file, large-file, merge-conflict, private-key, Codespell, and Typos checks; file-type-inapplicable hooks were skipped. The complete hook set could not finish because the repository-local ./gitleaks executable is absent. 4. End-to-end training, accuracy, and performance tests were not run locally. See merge request: Ascend/MindSpeed-LLM!5032 | 2 天前 | |
feat(mcore): add GLM-5.2 A5 POC scripts Co-authored-by: LinShua<707894133@qq.com> # message auto-generated for no-merge-commit merge: !5013 merge master_glm52_A5_sh into master feat(mcore): add GLM-5.2 A5 POC scripts Created-by: LinShua Commit-by: LinShua Merged-by: ascend-robot Description: Fixes #1825 ## What this PR does / why we need it? 1. Replaces the existing GLM-5.2 32B A5 POC entry with a 51B 4K FP8 configuration using TP1/PP2/EP4 parallelism, Transformer Engine MXFP8, index-topk=1024, fixed routing, the pipeline layout, and a matching log name. 2. Adds a GLM-5.2 210B 4K A5 FP8 POC entry with an eight-node TP1/PP2/EP32 configuration, Transformer Engine MXFP8 options, and portable data/tokenizer/checkpoint path placeholders. 3. Aligns --lr-warmup-iters to 500 in the GLM-5.2 744B A3 example, 744B A3 POC, and 210B A3 POC scripts. 4. This change is limited to five shell launch configurations and does not modify training runtime code or public APIs. ## Does this PR introduce any user-facing change? Yes. Users can run the new tests/poc/glm52/pretrain_glm52_210b_4k_A5_fp8_ptd.sh entry and the renamed tests/poc/glm52/pretrain_glm52_51b_4k_A5_ptd.sh entry. The 51B entry now uses the A5 MXFP8 and TP1/PP2/EP4 configuration, and the related A3 scripts now default to 500 learning-rate warmup iterations. ## How was this patch tested? 1. bash -n passed for all five changed shell scripts. 2. Changed-file pre-commit passed trailing-whitespace, end-of-file, added-large-file, merge-conflict, private-key, codespell, and typos checks. YAML, JSON, Python, and C/C++ hooks had no matching files. The repository-local gitleaks-offline-scan hook could not run because the required ./gitleaks executable is not present in the repository or local environment. 3. git diff --cached --check passed before amend. 4. The remote Git hook passed during git push --force-with-lease. 5. Functional, accuracy, performance, and end-to-end NPU training tests were not run. See merge request: Ascend/MindSpeed-LLM!5013 | 7 天前 |